The University of Wisconsin-Indiana University-National Optical Astronomy Observatories ('WIN') 3.5-m aperture telescope project's design concepts and development status are assessed. The WIN telescope employs a wide field of view in order to take advantage of recent advancements in multiobject fiber-optic spectroscopy. A novel support system is under development for the borosilicate honeycomb primary mirror blank which acts solely on the mirror's rear surface; mirror temperature will be actively controlled. The WIN telescope's control system will use a distributed, easily expanded and upgraded network of microprocessors connected to a master computer via serial bus.

Image-quality data are presented which were gathered at the University of Arizona's Multiple Mirror Telescope (MMT) using a very high quality 1.8-m mirror, for median image sizes of 0.72 arcsec in 8-sec integrations. The measured image is judged to have been degraded by 0.48 arcsec from that allowed by the free atmosphere by such factors as errors in telescope focus, collimation, mirror support, mirror seeing, and the combined optical figure of the five mirror surfaces required to bring the light to the image-analysis CCD. These data imply the appropriateness of a total optical error budget of 0.2 arcsec for 8-m diameter telescopes currently being designed.

The two 8-m aperture telescopes planned by the National Optical Astronomy Laboratories for Mauna Kea and Cerro Pachon will specialize in high-quality imaging in the visible and IR ranges. Spin-cast lightweight honeycomb monolithic mirror blanks will be employed, in conjunction with active primary support structure and thermal control. The focal-plane instrumentation will encompass a CCD imaging system, a multiple-aperture Cassegrain spectrograph, a multiobject spectrograph, high angular resolution IR imagers, a medium-resolution IR spectrograph, and a cryogenically cooled high resolution IR echelle spectrograph.

The British astronomical community is currently engaged in the development of an 8-m aperture visible/IR telecope whose structure is based on the successful Herschel 4.2-m telescope. The primary mirror will be of meniscus type, with a 40:1 aspect ratio and active support; two secondaries will be used, of which the first (f/7) will furnish a corrected 40-arcmin field, while the second (f/35) can be chopped for use in the thermal IR. Enclosure design options under consideration encompass a carousel, a conventional hemispherical dome, and a lightweight octagonal enclosure.

The Magellan Project has as its goal the construction of an 8-m aperture optical telescope at Las Campanas, Chile, whose f/1.2 parabolic primary mirror is of borosilicate honeycomb type. The principal configuration of the telescope will be with Cassegrain focus at f/6.26; image sizes are expected to be of the order of 0.25 arcsec rms over the entire field, from 0.33 to 1.10 microns without refocus. An IR Cassegrain focus is planned at f/15, with a chopping secondary of low emissivity. Despite the fast primary-mirror f-ratio, spherical aberration will be acceptably low when chopping is accomplished via rotation of the mirror about its vertex.

The W.M. Keck Observatory and its Ten-Meter Telescope are nearing completion at the summit of Mauna Kea. The 10-m diameter primary mirror has a 17.5-m focal length and is composed of 36 hexagonal segments. There will be seven Ritchey-Chretien f/15 foci: two of them at Nasmyth foci, one at Cassegrain focus, and four at bent Cassegrain foci on the elevation ring. There will also be an f/25 IR focal plane at the intersection of the optical and elevation axes, whose focus will be chopped by a beryllium secondary mirror. Image quality with a FWHM of the order of about 0.25 arcsec, and an 80-percent enclosed energy diameter of about 0.40 arcsec, are anticipated.

The Japanese National Large Telescope's 7.5-m diameter mirror will be of 20-cm thickness, thin-meniscus type. In conjunction with the use of a low thermal expansion coefficient material, this mirror concept is anticipated to have a lowest eigenfrequency of about 15 Hz; a floating support system will be employed which uses 264 contacts with the mirror. An altazimuth fork mount in conjunction with an oil mount-supported truss structure. The design of the structure is such as to allow ready access to the Cassegrain focus for polarimetric and IR bandpass studies. Pointing accuracy is projected to be better than 1 arcsec.

The ESO's Very Large Telescope will employ an array of four 8-m aperture telescopes which can be operated in either independent or combined modes, as well as in an interferometric mode. The primary mirrors used are of thin, Zerodur glass-ceramic meniscus type, figured for f/1.8; the overall design is optimized for f/15 at the Nasmyth foci. Active optics are employed to compensate for slowly varying deformation and thermal distortion effects, as well as wind buffeting. Wavefront sensing for active figure maintenance is accomplished by means of a Shack-Hartmann CCD wavefront sensor which is integrated with the imaging CCD used for field acquisition and tracking. Two types of enclosure are under consideration for the telescope which attempt to maximize natural ventilating flows.

Up to now telescope optics were usually specified in terms of geometrical errors which cannot be linked to the actual performance under atmospheric turbulence limitation. A more realistic approach is proposed which takes into account atmospheric seeing and diffraction. The main advantage of the method is that at the same time it describes the final performance of the telescope, and gives to the optical manufacturer the maximum freedom to define and possibly modify its own manufacturing error budget.

The Columbus Project's double 8-m diameter binocular (11.3-m aperture) telescope optical configuration encompasses (1) two Cassegrain foci at f/5.4 that are optimized for wide-field optical observations, (2) an additional two Cassegrain foci at f/15 which are optimized for the thermal IR, (3) a combined f/33 focus optimized for interferometry, and (4) several additional stations derived from either the folding or redirection of the first three types. Attention is presently given to the telescope's error budget, whose goal is the achievement of a wavefront structure function equivalent to images of 0.23 arcsec FWHM; the wavefront of the combined telescope, atmosphere, and instrumentation should be equivalent to a detected image of 0.34 arcsec FWHM.

The Very Large Telescope Interferometer (VLTI) is one of the operating modes of the VLT. In addition to consisting of the four stationary 8-meter-diameter telescopes, it includes a number of movable Auxiliary Telescopes which both complement the (u,v) plane coverage of the large telescopes and provide a powerful interferometric facility by itself (available 100 percent of the time). The current plans for the implementation of the VLTI are described. These plans will be finalized after the choice of the VLT site in 1990.

At the primary focus of JNLT, three-lens correctors with two aspheric surfaces will be used. They are designed for the wavelength region of 320-500 nm and 440-1100 nm, respectively, and provide images of less than 0.2 arcsec rms radius for a field of 30 arcmin diameter. The corrector for the blue region will include an atmospheric dispersion corrector. Characteristics of aberrations that arises in JNLT primary corrector are discussed.

This report considers the optics of single versatile telescopes, rather than MMTs and interferometers. There are many new points to consider in the optical design which make the problems of the eight-meter different from those of previous telescopes. A wide-field focus is required at 6-8 ft, but should not conflict with the need to switch between Cassegrain and Nasmyth positions; this can lead to a new type of design. It is not necessary to use either of the textbook optical systems, Cassegrain or Ritchey-Chretien (R-C), at the f-number, but all possibilities must be considered. The present paper describes detailed possibilities for systems of the R-C type, and further work in progress on conventional Cassegrain and intermediate systems is outlined for completeness and comparison.

The adaptive optics techniques currently under development for telescopic mirrors of 8-m aperture class are presently projected to serve as the bases for mirrors of the order of 32-m diameter, in virtue of the rapid increase in the sensitivity advantage of diffraction-limited imaging with increasing aperture size. If the technology of atmospheric wavefront error sensing by means of artificial starlight is perfected, diffraction-limited imaging will be possible over the entire sky at even optical wavelengths. The corresponding resolution of 0.004 arcsec, in conjunction with the light-gathering power of a 32-m aperture, will constitute a milestone advancement for astronomical research prospects.

The application of full adaptive optics to astronomical telescopes in the foreseeable future is likely to be limited to infrared wavelengths greater than 1 micron both because of the limited number of bright enough wavefront sensing objects at visible wavelengths and because of the complexity and expense of making an adaptive optics system for large telescopes with the large number of elements required at visible wavelengths. Adaptive optics designed for infrared wavelengths do, however, improve the image quality at wavelengths shorter than the design wavelength, thus improving the sensitivity of interferometric imaging at those wavelengths.

It is generally thought that the resolution of large ground-based telescopes is limited by atmospheric turbulence rather than by diffraction from the telescope aperture. However, longer wavelengths are less affected by atmospheric turbulence than shorter wavelengths and, conversely, longer wavelengths are more affected by diffraction from the telescope aperture. An optimum wavelength exists where these two counteracting effects balance. At this wavelength, maximum (diffraction-limited) resolution is obtained. In night seeing conditions at typical telescope sites, the optimum wavelength is in the range 1-2.5 microns. For a 5-m telescope, it should be possible to obtain resolution of the order 0.05-0.15 arcsec routinely at these wavelengths. However, to facilitate such precise resolution the telescope must be diffraction-limited.

As applied to the 4.2-m William Herschel Telescope, the multiaperture real-time image-normalization system presented implies a wavefront whose size requires a mask of six optimally-scaled subapertures. These subaperture images are separated and examined on a single image photon detector which yields x, y, and t coordinates for each recorded photon. The motions of these images feed back to six independent piezoactuated active mirrors which act to null the image motions at a CCD focus. Data are presented from two image normalization runs, with and without active mirrors, which illustrate the size and variation behavior of the coherent seeing length, characteristic seeing times, and power spectra.

Since liquid mirrors are potentially useful in science (e.g., astronomy, atmospheric sciences, and optical testing), work has been undertaken to determine whether they are technologically feasible. A testing tower has been equipped with a scatterplate interferometer interfaced with a CCD for data acquisition and a microcomputer for data analysis. This equipment was used to test a 1.5-m-diameter f/2 liquid mirror, showing that it is diffraction limited; interferometric measurements give Strehl ratios of order 0.8. A 2.7-m-diameter liquid mirror and astronomical observatory presently under construction is briefly described.

General design considerations of objective-mirror coronagraphs are presented. A 1-m-focal-length prototype reflecting coronagraph based on a 5.5-cm aperture spherical superpolished silicon mirror objective is described. The design is simple off-axis reflection from the objective to a conventional coronagraph optical system (occulting disk, field lens, Lyot stop, and imaging system). This instrument has produced the first images of the emission corona using a ground-based reflecting coronagraph. A second prototype instrument based on a 15-cm aperture superpolished fused-silica mirror is also described.

The design concept of a wide-field astronomical imaging telescope for use as a payload on an unmanned space platform, a Space Station attached payload, or a Delta-class Explorer is described. The instrument is based on a space Schmidt telescope concept studied by NASA and ESA (1979) for Spacelab missions. The astrophysical objectives include all-sky surveys in the UV and NIR ranges. Objects of interest include very hot and very cool stars and the interstellar medium. The UV range is inaccessible from the ground, and large-area surveys and sensitive imagery of diffuse sources are impractical with current or planned UV space telescopes. The NIR range is severely compromised in ground-based observations, particularly of diffuse sources, by airglow emissions, and no wide-field NIR space telescopes are currently approved for flight.

A 1.6-meter diameter f/0.95 all-reflecting telescope was designed to observe orbital debris particles as small as 1 mm from the shuttle payload bay. The telescope was specified to have a flat focal surface without the imposition of refractive elements. Two design configurations involving three mirrors were evaluated - a reflective Schmidt-Cassegrain and a modified Paul corrector. The Paul system was found to be more compact and appropriate for this application.

The 2dF project aims at giving the AAT Prime-focus a 2-deg diameter corrected field of view geared primarily toward multiobject fiber spectroscopy. The combination of the 4-meter aperture of the AAT, a 400-fiber autopositioner and high throughput spectrographs should make the 2dF a powerful facility for statistical spectroscopic studies in the years to come. The design study for the corrector, fiber-positioner and spectrographs is now complete and the main conclusions are reported.

It is very significant to develop the fiber-optic telescope that makes use of the large cross-section (i.e., large information) image-transmitting bundle and to realize the system combined 'rigid' with 'soft', in order to achieve the observation of remote objects in some difficult conditions. This paper discusses the key technical problems by which simpler structure, higher performance indices (magnification, field of view) of systems, finer resolution and better results of observation are realized under the condition of the definite image-transmitting bundle. These problems are: to choose and use high performance LCSITB; to select a reasonable eyepiece to restrain the mosaic pattern of the image; to make use of the objective lens that has long focal distance and large aperture to obtain optimal coupling between the objective lens and LCSITB and to obtain the higher resolution and sufficient brightness of the image; and to select the simplest erecting system. Using the fiber-optic telescope which is designed and developed, it is observed and photos show the remote objects from about one kilometer and good results are achieved.

An improved chopping secondary system for a 300 mm balloon-borne IR telescope (BIT) has been developed. BIT is planned to be flown for the first tine in May 1990 in China. Design and overall properties of the chopping secondary system are presented. The important feature of this chopper is its control driver, which is a new type of vibrator. The whole system including the control circuit is very simple, lightweight, and low-cost. A 2-msec rise time of the 93 mm secondary mirror motion was achieved at a frequency of 10 Hz.

In order to achieve a high speed/high precision/low power consumption IR telescope chopping system which excites no mechanical resonances, in conjunction with a device size smaller than the secondary mirror diameter, a chopping servo has been developed which uses feedforward control with optimized driving torque as well as feedback control guided by robust stability theory. The compactness of the device has been achieved through the use of an Nd-Fe-B magnet motor and the balancing of the mirror against the mass of the electromechanical system. The chopper has a 10-msec transient time, and stability in excess of 1 arcsec, for square-wave chopping with amplitudes of as much as 300 arcsec.

Telescopes designed for non-conventional imaging of near-earth satellites must follow a unique set of design rules. Costs must be reduced substantially and the design must accommodate a technique to circumvent the atmospheric distortions of the image. Apertures to 12 meters and beyond are required along with alt-alt mounts providing high tracking rates. A novel design for such a telescope has been generated which is optimized for speckle imaging. Its mount closely resembles a radar mount and it does not employ the conventional dome. Costs for this design are projected to be considerably reduced compared to conventional designs. Results of a detailed design study will be presented. Applications to astronomy will be discussed.

The factors contributing to fringe contrast decrease for telescopes working near their diffraction limits are summarized. These factors include variations with time, such as atmospheric variations, vibrations, pathlength drift, and fringe tracking noise; and variations accross the pupil; variation with wavelength and factors relating to polarization effects; unequal beam intensities; detector resolution; and pupil transfer geometry. The effects on multispeckle images are also considered. The resulting error budget for the Very Large Telescope Interferometer (VLTI) is derived. It is concluded that the total random error in the fringe contrast is 4.2. The total calibratable systematic error amounts to 34 percent (27 percent due to the instrument, 9 percent due to the atmosphere).

The development of generic optical designs for supporting wide-field imagery from an array of individual telescopes is presented. These systems will be employed to gather imagery over a total field of 0.5 to 1.0 degree. Various design possibilities are studied assuming the use of a square field, with the option of utilizing a slit field. A general optical design concept that can be transferred to a reasonable number of fairly closely packed arrays is examined, evaluating the promising three- or five-mirror arrangements.

This paper discusses the requirements posed on the ESO Very Large Telescope (VLT) Interferometer by the applications that require a field-of-view larger than the Airy disk of the individual telescopes. The most essential requirement for such wide field-of-view use of interferometric arrays is the maintenance of the pupil configuration, which applies to all the details of this configuration. Not meeting this requirement leads to path-length differences among the rays of each of the telescopes composing the array. An error budget for the optical design parameters of the VLT Interferometer is derived.

The sensitivity of image quality to various system and subsystem parameters has been studied in order to determine the utility of imaging phased telescope arrays to wide field of view (FOV) applications. An error budget tree is developed to include optical design errors, assembly and alignment errors, optical fabrication errors, and environmental errors. Trade studies, parametric analyses, and previous engineering experience permitted the derivation of design and engineering tolerances from error budget allocations based on known state-of-the-art performance characteristics. The FOV limitations of the residual optical design errors (off-axis aberrations) are investigated in detail. It is shown that the somewhat benign (for conventional optical systems) aberration of field curvature results in field-dependent relative phase (piston) and pointing (tilt) errors, which rapidly degrades the image quality of phased telescope arrays as the FOV is increased. Thus extremely light tolerances on residual field curvature are needed for telescope diameters larger than one meter.

A new geometric and physical optics simulation code for predicting the performance of phased array telescopes is described. A skew aspheric ray trace routine computes a composite spot diagram and an array of optical path differences for the entire telescope system. This includes the four nearly identical afocal telescopes and the single combiner telescope. A second routine then computes subaperture Zernike aberration coefficients. A wave optics code computes point spread functions, and a final code computes optical transfer functions. The simulated performance of Air Force's Multipurpose Multiple Telescope Testbed (MMTT) is then presented and discussed. All optical surfaces of the telescope plus in situ measured aberrations are simulated. The results show that the telescope is nearly diffraction limited at small field angles, but suffers from phase and tilt differences among the telescopes at field angles above two milliradians.

A key feature of the Multipurpose Multiple Telescope Testbed (MMTT) is its relatively wide field of view -- up to 30 arcminutes total. A thorough evaluation of the telescope array necessitates some form of image analysis over this field. System designers chose the star test, here modified to simltaneously display point spread functions (PSFs) at several locations in the image plane. Deviations of the image structure from ideal indicate control system deficiencies. Diagnostics hardware was designed and built to both simulate numerous unresolved light sources throughout the field of view and to display several of the corresponding images for detailed analysis. The resultant multiple target generator is an array of precision pinholes complete with beam-converting optics that require minimal alignment. PSFs are displayed with modern solid-state video equipment. Image irradiance cross-sections can be readily compared with theoretical predictions.

A phased-array telescope is formed by combining a number of optical telescopes to yield the resolution of a single much larger telescope. One such telescope, the Multipurpose Multiple Telescope Testbed (MMTT), consists of four 20-cm-aperture telescopes phased together with a 30-arcmin field of view. This paper illustrates the model identification of the MMTT system. Each of the 4 channels of the system consists of over 30 subsystems and/or components. Various experimental tests were set up for the MMTT components using a white Gaussian noise source on a spectrum analyzer. The input/output signals for each experimental set up were measured and an ASCII file was created on a personal computer. The transfer function of each subsystem was identified using either the spectrum analyzer and/or through standard system identification software on an AT-class PC. The model thus identified can be used to study system's behavior by simulation as well as designing various controllers for tilt, piston, and pupil geometry control purposes.

The control system for the Multipurpose Multiple Telescope Testbed (MMTT) is described. Performance of the control system is measured against the goal of maintaining lateral pupil geometry to within one micron of the ideal. The MMTT control system consists of four channels. A discrete-time-varying Kalman filter processes the sensor measurements to estimate lateral pupil geometry and piston errors in each channel. The controller integrates the Kalman filter estimates along with the x- and y-tilt measurements, which are treated deterministically. Multiplication by an axis separator matrix converts the control input commands into appropriate piezoelectric transducer actuator commands. The commands from each control channel remove piston, tilt, and pupil geometry error between each telescope and the telescope denoted the reference. The entire control algorithm is implemented on high-speed array processor boards. Preliminary test results show that the control system is accurately controlling lateral pupil geometry for small field angles. For larger field angles, the system is unstable due to parameter variations.

The Multipurpose Multiple Telescope Testbed is described, and initial tests are discussed. After the optical quality of individual telescopes was established with interferometric tests, the cophasing and image superpositioning accuracy of the array were measured using star tests. Point spread functions were calculated with a physical optics code. Preliminary star tests using two of the four telescopes are presented and compared with the predicted pattern. It is concluded that the preliminary results obtained lend credence to the given method of controlling lateral pupil geometry and to previous calculations of optical aberration and optical alignment tolerances.

A new wavefront sensing technique called curvature sensing is described. It maps the wavefront total curvature rather than its slope and has been applied to an experimental seeing monitor which detects turbulence induced fast focus fluctuations. Some of the advantages this monitor presents, as compared to DIMM's, are: (1) sensitivity is increased by the use of a circular pupil, (2) the cost is lowered by the use of a photomultiplier, (3) the loss of signal is prevented by the system's fast run, (4) the system runs continuously, and (5) the noise bias is continuously measured and subtracted out.

Astronomical observatory site testing programs in the USA and USSR have used a variety of stellar image motion monitors in the selection of the best sites for the construction of large (6 to 10 meter) telescopes. While there appears to be a reasonable agreement between microthermal and sodar results for the better sites in both countries, there remain unexplained inconsistencies in measured seeing, especially at Mauna Kea, Hawaii and Mount Sanglok. The photoelectric seeing monitor built by Scheglov (1984) of the Moscow Sternberg Institute, and the National Optical Astronomy Observatories site-survey intensified CID seeing monitor have been mounted on the same telescope. Simultaneous image motion data recorded are

During two short campaigns intensive coordinated measurements have been performed to determine the various contributions to image degradation on Mauna Kea. Some of the results already obtained are presented here.

Interferograms of the front surface of a zenith-pointing 1.8 m mirror for a range of surface-to-air temperature differences were analyzed using two different methods to show that significant mirror seeing effects do not occur in the range of + or - 0.5 C, and may be acceptably small up to + or - 1 C. Mirror seeing was shown to comply with Kolmolgorov turbulence laws up to limits that are apparently due to the test setup enclosure. Small seeing effects were discernible despite much larger image error sources.

The project to enlarge the Multiple Mirror Telescope (MMT) to a 6.5 m single primary mirror telescope is described. The goal is to provide a telescope which is competitive with the existing MMT in tracking and pointing performance (0.2 and 1.0 arcseconds, respectively) but has more than twice the light gathering power and 15 times the angular field of view. The existing mount and building will be used with minor modifications so that the cost of the project is relatively modest. Casting of the 6.5 m mirror is scheculed in early 1991 and first light in late 1993.

The seeing degradation in the Japanese National Large Telescope (JNLT) caused by its hemispherical dome is investigated. A possible plan of the JNLT site layout and the thermal control concept are introduced in order to attempt to reduce the seeing degradation induced by the dome to the 0.1 arcsec FWHM range budgeted. A three-dimensioal compressible fluid analysis of the inside and outside the dome, including heat transfer effect, is developed and used to understand the seeing degradation mechanism as well as the wind buffet effect on the telescope and the primary mirror.

A site testing telescope (STT) was placed for a period of 3 months outside the building, and for a period of 4 months inside the building of the Multiple Mirror Telescope (MMT) located on Mt. Hopkins, and measurements of the astronomical seeing were carried out with both the STT and MMT. A comparison of the simultaneous and interleaved measurements with the two telescopes reveals a tight correlation and a well-defined relationship between the seeing image sizes determined with each telescope. The STT predicts very well the size of a long-exposure image obtained with the MMT. There exists for the MMT an optical blur component of about 0.47 arcsec that is revealed in image size data obtained with only the MMT and that is also revealed and must be accounted for in comparisons between the STT and MMT. Also discovered is a downward component of seeing that is caused by a trailing downwind plume of cold turbulent air that is shed off the radiatively cooled exterior building surfaces. The interior dome component of seeing is remarkably small (upper limit about 0.2 arcsec). The median site seeing is determined to be 0.55 arcsec at an effective wavelength of 7165 A for the MMT observations, or about 0.59 arcsec at 5000 A. The 10 percentile value is about 0.28 arcsec (0.30 arcsec at 5000 A). The median seeing observed with the MMT from one and a half years of data is 0.72 arcsec (0.75 arcsec at 5000 A), and is degraded from the site value virtually entirely by the optical blur component.

Results are presented on the evaluation of the site for the Nordic Optical Telescope (NOT) at its site at the Roque de los Muchachos Observatory. Basic design features of the NOT are described together with the parameters that define image quality, with special consideration given to the role of atmospheric turbulence and the factors taken into account in the selection of the telescope site. Attention is given to the optical elements of the NOT and to its mechanical structure as well as to the thermal control of the telescope, the enclosure, and the ancillary instrumentation. Results from first observations at the NOT point to excellent observing conditions in terms of transparency and extinction stability as well as image quality.

A 3.5 m telescope is under construction at Apache Point near Alamogordo, New Mexico, at an elevation of 2800 m. A thermal model of a telescope enclosure is described. The model evaluates various strategies for minimizing local sources of image degradation (dome seeing). Direct and diffuse insolation, radiation to the sky, conduction, and the thermal inertia of the walls, interior air, roof, and structural steel are included. It is observed that highly reflective surface coatings reduce heat absorbed during the day, but are not very effective in reducing heat transfer in the telescope chamber at night, assuming that components with large heat capacities or thermal time constants are insulated.

The preservation of image quality is one of the most important design criteria for a telescope enclosure. Image quality is adversely affected by the presence of fluctuations of air temperature in the light path. In an ideal telescope enclosure, the air inside the telescope chamber is continually replaced by ambient air before it has the opportunity to be warmed or cooled by the surfaces of the telescope enclosure. In the present study, dye in a water tunnel is used to trace flow around and through four models of 8 m telescope enclosures. The models are constructed of transparent acrylic and are 1:172 scale. Flow attributes are compared for the four enclosure designs under consideration. The enclosure models are oriented at various angles with respect to the flow direction. Additionally, each design contains apertures in the exterior walls which can be opened to improve flushing and other flow characteristics. For the octagonal and rectangular designs, flushing of the telescope chamber through the open vents is good in all orientations. Flushing of the cylindrical concept, which has no vents in its side walls, is poor when the slit is oriented from 60 to 120 deg with respect to the flow direction. Flushing of the hemispherical enclosure is poor at angles larger than 60 deg. With the vents closed, flushing is poor for all designs, especially at angles greater than 60 deg.

The cost of an 8-m telescope is still too high for most national observatories. Cost reductions must involve new technology for the primary mirror: its material and figuring, its mass, handling and the aluminizing plant. The present scheme addresses these problems using some features of the Keck technology, but is simplified. Consideration is given to a segmented mirror in which the segments are radially cut sectors. The sectors for an 8-m aperture will fit inside a 4-m aluminizing plant (which already exists on some sites, and in any event is cheaper than an 8-m plant). Zero-expansion material (glass-ceramic or fused silica) may be used. The model thickness is about 10 cm. The proposed method of production is to figure the whole set assembled as a single mirror. No cutting takes place after figuring.

The fabrication of high aspect ratio mirrors up to 8-m in diameter presents numerous new challenges in the field of optical manufacturing technology. These mirrors, with aspect ratios in the region of 40:1, represent a new class of optics which requires unique handing, figuring, polishing and metrology approaches. A facility conceived explicitly for the manufacture of high-quality large-aperture primary mirrors for astronomical telescopes is described.

While coefficient of thermal expansion (CTE) differences between the facets of a large (8-m diameter) telescope's hexagonal-mosaic mirror generate optically significant deformations in the midspatial frequency regime, when the average temperature of the mirror undergoes substantial change, it is presently established that the magnitude of these deformations lie within acceptable limits for the active figure-control system envisioned for such telescopes. It is nevertheless recommended that CTE discontinuities within such mosaics be consciously planned for at the outset of construction planning.

The National Optical Astronomy Observatories (NOAO) have been working for several years to develop the technology for 8-m telescopes using structured borosilicate glass primary mirrors. In March 1989 the final stage in this technology development program began with the delivery of a 3.5-m mirror blank, cast under NOAO contract at the University of Arizona Mirror Lab. The project will have four phases: (1) initial fabrication, (2) testing of support and thermal systems, (3) aspherizing the mirror and rework of the support and thermal systems, and (4) final acceptance test. At the conclusion of this effort there will be a finished 3.5-m f/1.75 mirror, which will then become the heart of the new WIN telescope on Kitt Peak.

It is presently noted that the availability of CCD cameras, in conjunction with state-of-the-art computer hardware and software, render Hartmann testing in optical shops a viable alternative to, or checking procedure for, other types of tests. An evaluation is made of data reduction methods for this test. While the reduction of data-fitting to derivatives of Zernike polynomials filters out high-frequency surface information, and runs the risk of yielding incorrect coefficients, it may be able to find special-purpose applications. The least-square integration of a surface is reliable, but time-consuming in the case of large data sets. Subaperture Hartmann results can be assembled to obtain the entire surface; rastering can double the high spatial frequency spatial information of the surface.

A modification to the usual phase-shifting interferometry algorithm permits measurements to be taken fast enough to essentially freeze out vibrations. Only two interferograms are time critical in this 2 + 1 algorithm; the third is a null. An error analysis has been performed for this new algorithm. The implemented system acquires the two time-critical interferograms with a 1-msec separation, on either side of the interline transfer of a standard CCD video camera, resulting in a reduction in sensitivity to vibration of one to two orders of magnitude. The required phase shift is achieved via frequency shifting.

In order to reduce polishing costs and correct unexpected errors in fabrication and polishing, the support of very large optics can be actively enlisted in telescope mirror optical figure adjustment. A set of leaf springs is used by the Keck Ten-Meter Telescope to apply moments about the pivots of the mirror mosaics' whiffletree support. The springs successfully reduce the polished rms surface error by a factor of 6 to 15, while reducing the 80-percent enclosed energy diameter by a factor of 2.5-6.0. Additional current limitations on figure improvement include the difficulties of polishing higher spatial frequencies and predicting warping during mirror fabrication.

We are in the process of polishing a 1.8-rn f/i ellipsoid with an actively stressed lap. As a preliminary exercise, we have polished the mirror as a sphere using a rigid subdiameter lap. The overall surface error was 25 nm rms, and the surface met a specification corresponding to i/8-arcsec image quality. A stressed lap 600 mm in diameter was designed and built to polish the mirror as an f/i ellipsoid. It consists of an aluminum disk which changes shape continuously under the influence of 12 moment-generating actuators. These actuators are programmed to produce the shape changes necessary to make the lap fit the mirror surface as it moves across that surface and rotates. In this paper we describe the principles and design of the lap, test results, and progress to date in polishing the 1.8-rn mirror.

Technology developed for computer-controlled mirror manufacturing is described. The polishing tool is equipped with electromagnetic force actuators to regulate the local polishing forces according to the measured mirror errors. Optical testing with the interferometric modification of the Hartmann test and a CCD-camera as a detector allow accurate and fast measurement with high sampling frequency. Air turbulence and vibration effects are minimized in the workshop which is blasted into the bedrock and equipped with good thermal insulation.

The numerical control device for zonal figuring of aspherical surface according to polar coordinates has the advantage of simplicity and compactness in structure. The method to predict the surface profile after figuring is presented. Effort has been made to develop some methods for automatic generation of the zonal figuring procedure by computer. An example of making a Schmidt correcting plate by zonal figuring is given. The results of various figuring procedures are compared and discussed.

A composite mirror consisting of a carbon fibre reinforced fused quartz(CFRFQ) substrate and a fused quartz mirror surface which could be polished to a high precision on the surface figure was prepared. The experimental results indicate that the randomly oriented CFRFQ with no matrix cracking due to thermal expansion mismatch have a improved mechanical and thermal endurance and the interfacical bonding between the substrate and the fused quartz mirror is very densification. It is estimathd that this type of composite mirror would be a potential material for future space telescope.

Reflectors for antennas using fiber-composite technology today represent the state of the art, and mirrors for radio telescopes occasionally are already made using this technology. For even shorter wavelengths, including the visible light, glass mirrors have been used almost exclusively and, rarely, metal mirrors (for example made of beryllium). In general a surface contour accuracy of about 1/20 wavelength is required. Years of development work based on experience gained in aircraft construction and space technology made it possible to improve fiber-composite technology to such an extent that a substitute for heavy glass has become feasible, opening up new applications for lighter and/or larger mirrors.

The objective of x-ray imaging telescope to be used for observing flares and sun- spOt8 of the sun, which is aboard on astronomic satellite, is developed accoding to chinese space science developmerm plan, it consists or two aspherical barrels which is made from metal material. The requirement of accuracy for surface fiqure of the mi- rror is A/5, surface roughness is ioA RMS, the roundness of inner diameter is 0.5 ,m. The coinbinatory barrel is made of aluminium alloy, plating electronless nickel on ml- rror surface after machining, and then doing superprecision processing by superpreci- sion diamondturningmachine tool, finally doing grinding and polishing. In order to provent deformation of workpice from mounting and clamping, it would effect the ac- curacy of surface figure, so we adopted the method of floating grinding and polishing. The paper related mechanical and optical processing technologies of the barrel, and facilities, and went into details on supersmooth polishing processing s well as the testing methods of the accuracy of surface figure and surface roughness, Final mea- cured results indicate clearly the geomtric precision of combinatory barrel, and sur- face figure, and surface roughness of the mirror have achieved design requirements el- ementaly.

For about 20 years SCHOTT has been supplying the glass ceramic ZERODUR, a material with a thermal expansion of virtually zero. ZERODUR is excellently suited for the manufacture of mirror substrates for astronomical telescopes. SCHOTT has accumulated substantial experience in melting and processing of this material. New techniques, such as spin-casting, have been developed by SCHOTT for the manufacture of thin monolithic mirror blanks exceeding 8 m in diameter. These developments resulted in SCHOTT obtaining the order for the manufacture of 4 meniscus shaped shells for the Very Large Telescope of the European Southern Observatory. The corresponding production facilities are currently being erected with the first casting scheduled for the end of 1990. SCHOTT has also performed considerable developmental work in the area of lightweighted ZERODUR mirror substrates which can be fabricated by applying one or a combination of the following techniques: forming of the lightweighted structure during casting, fusion of individual components to a total structure and lightweighting of a massive block by various machining methods.

The purpose of this paper is to help determine whether or not a lightweight mirror is an appropriate engineering solution in place of a more traditional, solid mirror. The relatively recent manufacture of borosilicate lightweight mirrors both by casting1 and by the Gas Fusion process' make this study timely. Because lightweight borosilicate mirror substrates tend to. be substantially less expensive than the equivalent lightweight substrates made of very low expansion materials, a review of the properties of lightweight mirror structures in general will make it possible to determine if the cost savings of the borosilicate substrates are justified when balanced against the performance of the very low expansion mirror substrate materials.

ULETM, a zero expansion glass, offers many advantages as mirror blank material due to its thermal and mechanical properties as well as the flexibility it offers in design and fabrication. Produced by the flame hydrolysis process, this titanium silicate has high homogeneity of Coefficient of Thermal Expansion (CTE) that can be determined within any piece of ULE by nondestructive ultrasonic measurements. The ability to fusion-seal the glass while maintaining strength offers mirror manufacturing design freedom. Coming has produced a 12 inch square plano blank and a 24 inch diameter meniscus, both fabricated with cooling channels, using existing fusion techniques. The square blank was manufactured by actually machining grooves into the plates prior to fusing them together. The meniscus production consisted of fusing bars of ULE between the two outside plates. The meniscus was subsequently thermally tested by NOAO. A 4 meter solid monolithic meniscus with cooling channels is an extension of this fusion technology.

There is considerable interest and excitement in the astronomical community about the near-term realization of 8 to 10 meter class telescopes. In this paper we will describe the design of a polishing shop uniquely configured to acconmiodate these mirrors. Simply stated, these mirrors are very large, very heavy, and very flexible compared to existing telescope designs. Large implies long path length metrology and the control of vibration as well as the need for vacuum to eliminate air turbulence. It also implies a significant investment of time and money and therefore, safety is a paramount design parameter. Heavy introduces in-process handling concerns, an always present safety issue. Flexibility, whether a thin solid or structured approach is employed for the mirror blank, requires that some form of active shape control be employed to maintain figure in the presence of gravity. This can be exploited to relax the figure requirements in the long spatial period regime which has significant implications on the type of machinery that will be needed to figure and polish the optic. We will describe in this paper how the facility design addresses these issues.

Air flow tests with a 61-cm diameter, 8.9-cm thick mirror have indicated that the circulation of ambient temperature air through the channels at 10 m/sec sufficed to maintain the mirror's front surface below the critical temperatures that would result in good mirror seeing; the effective front-surface thermal time constant for these conditions was 1 hr. An extrapolation of these results to mirrors of greater thickness, without channels, indicated that the use of a heat exchanger to control rear surface air temperature could be as effective as ambient air flow through internal channels, up to mirror thicknesses of about 10 cm.

A thermal sensor measurement system, with a resolution better than 0.01 C has been built for the NOAO 3.5-m mirror project. A total of 1000 low-cost precision temperature current transducers are multiplexed through a single 16-bit ADC so that temperature profiles may be obtained in less than 30 sec. The problems and pitfalls encountered in taking a device that was designed to be accurate to 0.5 C and obtaining reliable measurements to + or - 0.003 C with a large number of sensors are discussed.

The Astrophysical Research Consortium, the Magellan Project, and the National Optical Astronomy Observatories are collaborating in an experiment to study temperature control of a honeycomb mirror in an observatory environment. A full-scale mockup of a 60-deg sector of a 3.5-m borosilicate honeycomb mirror was constructed of glass and installed in the Apache Point Observatory (APO) 3.5-m telescope mount. The response of this mockup (the wedge) to diurnal temperature variations was monitored by measuring the glass temperature at about 300 locations distributed throughout its volume. The temperature of the wedge can be controlled via air circulated through its interior.

The Steward Observatory Mirror Laboratory has implemented a large borosilicate honeycomb telescope mirror polishing system in which thermal distortion is reduced to negligible levels by maintaining the glass in an isothermal state to within 0.1 C. Testing of the polished surface is conducted in air, using a laser and interferometer mounted above the mirror; the control of refractive index variation in the laser's light-path entails that the air also be isothermal, to within 0.2 C. Thermocouples are used as sensors in the polishing room, in air ducts, and on the mirror. Measurements are made to an accuracy of 0.005 C at the rate of one thermocouple/sec.

The tracking performance of advanced technology telescopes is presently predicted by a time-domain nonlinear control model which incorporates the complex frequency-dependent transfer characteristics of a type II servosystem, including (1) rate and acceleration feedforward, (2) gimbal-drive motors, (3) motor power amplifiers, (4) mechanical drivetrain, (5) telescope structure, and (6) encoders. Disturbances generated by bearing friction, drive motor magnetic cogging, drive motor friction and torque constant variations, wind loads, etc, are included to enhance the accuracy of tracking error predictions under operating conditions. The model is useful in both initial design studies and the evaluation of proposed design modifications.

Instrumentation and procedures developed for the study of friction-roller slippage in off-axis couplings of telescope position encoders are presented, and the designs of rollers and their mountings are examined with a view to the characterization of adjustment techniques yielding optimal alignment of the rollers to minimize slippage. Attention is given to the use of various types of fiducials, including GEO satellites and inexpensive optoelectronic sensors, to gather measures of both telescope-position encoder repeatability and telescope hysteresis which are independent of the telescope-pointing model.

An Arecibo radio telescope-like optical telescope is under development which will operate in semitransit mode for spectroscopic survey work, using a five-axis servomechanism in the telescope's upper section to track an object's image locus on the focal sphere. Once the telescope is in the requisite azimuth position, a transit time for each object relative to the telescope structure can be determined and a strategy devised for driving the fiber-carrying optical head into conjunction with the image focus. The trigonometric mapping between the hour angle and declination functions of each object, on the one hand, and the five coordinates of the tracking device in the telescope's upper section, on the other, is a special transformation applicable to all telescopes of this type.

To take full advantage of the superb seeing available at the best sites, it is essential to control the attitude of a telescope with the utmost accuracy. Conventionally, this entails supporting the optical elements with as stiff a structure as is feasible and moving the structure with as precise and as smooth a drive system as can be afforded, using a massive pier as the datum for angular measurement and for reacting drive torques. The problems and costs in this approach increase steeply with telescope size, especially if the requirements for minimizing local seeing degradation lead to the telescope being mounted high above the ground or having minimal protection from wind. Ground transmitted vibration may also be a significant problem on some sites. A better approach is to provide the higher frequency angular corrections by supplementary drives which react against angular inertias, analogous to the reaction wheels used for pointing spacecraft. Then, attaining fast correction of errors, such as those induced by wind, does not require particularly stiff structures and drives between the optical assembly and earth. In fact these links with the earth might, with advantage, be made deliberately compliant to isolate the optics from ground vibration. If, in addition, the higher frequency angular errors are sensed relative to an inertial frame, the need for precision gearing is greatly reduced. Computer simulations indicate that the inertial drive technique is quite practical. For very large telescopes, this approach offers better performance at substantially lower overall cost than when conventional drives are used alone.

The Apache Point Observatory 3.5-m telescope, loated in the Sacramento Mountains of New Mexico at an elevation of 2.8 km, is awaiting the installation of the primary mirror. The design of this telescope includes a number of innovative features including a lightweight honeycomb borosilicate mirror, friction drives and encoding, and complete computer control to facilitate remote use. A 0.3-m telescope was attached to the telescope mount to monitor tracking and pointing performance. Currently, the telescope tracks open-loop to 0.3 arcsec rms over 10 min and points to 5 arcsec rms. Improvements to the servosystem and encoder mounting should make it possible to meet the goals of tracking to 0.2 arcsec rms and pointing to 1 arcsec rms.

The present evaluation of the design philosophy of the Apache Point Observatory's 3.5-m altazimuth-fork mount telescope gives attention to such detail-design problems as the mirror-cover system, the upper azimuth bearing and support system, and the spherical bearing mountings. A four-roller constraint for the upper azimuth bearing and support system proved to be a superior design solution than the originally intended two-roller constraint. A cone-shaped structure on the lower fork defines the azimuth axis; the center of the drive disk defines the top end of the axis while the spherical roller thrust bearing defines the lower end.

A 16 meter optical telescope is described. The telescope tube has a large and heavy truss that creates deformation; when the telescope tube is rotating, four servos push forward or pull back the main mirror truss to automatically compensate for the deformation. To obtain high image quality and tracking and pointing accuracy, dynamic deflection must be measured and the measuring value should be fed back to servo motors to provide automatic compensation until the system deflection is zero. The interferometer of the automatic compensation system is fixed at the rear of the main mirror to avoid influencing the main optical path. Details are discussed on the measurement of the relative deflection, the honeycomb mirror, and a scheme for main mirror measurement.

The Japanese National Large Telescope (JNLT) will use a thin meniscus mirror of 7.5 m diameter and 0.2 m thickness as a primary mirror. One of the biggest concerns is mirror deflection due to dome-inside wind loading, which varies with time not only in its pressure amplitude but also in its pressure profile on the mirror surface. To analyze such mirror deflection, a method combining modal analysis and random vibration analysis was introduced. From the rms mirror deflection obtained by this method, the image quality degradation is considered to be better than that budgeted as wind loading degradation. Information about the method and typical result of the analysis are presented.

The reasons are presented for the high performance of the Columbus Project Telescope whose design is based on two short focal mirrors, large drive and support radii, and a short load path to the ground. The radius squared is argued to be the most important tool for the inprovement of performance in the large optical telescopes, since the stiffness of the mechanics is proportional to the radius at which they act squared. The six Columbus telescope finite element models show that optimization of a structure depends more on the initial conditions (radius squared) than the truss shape or elements. It is concluded that future exploitation of radius squared could lead to higher performance for the very large telescope.

The Universities of Texas and Penn State are working together on an Arecibo-type optical telescope to be utilized in a semitransit mode for spectroscopic survey work. Its optics include a spherical primary mirror, a 2-element all-reflecting Gregorian spherical aberration corrector, and a series of optical fibers that will transmit light to a family of spectrographs. An optical support structure is being developed to permit position adjustment in azimuth only. During an azimuth position change, the instrument's entire weight is borne by steel rollers bearing on a circular crane rail of standard section, with support loads transmitted to the telescope base through pneumatic springs. Extensive application of various analytical procedures and computer-aided engineering tools has effectively allowed the detailed examination of several design iterations, thereby increasing the probability of success in the realized structure.

The support system of the 8.2 meter primary mirrors of the ESO Very Large Telescope consists of 150 axial support at the back surface of the mirror and appr. 60 lateral supports at the outer and inner edge of the mirror. This paper describes the general design of the support system and the prototypes of the axial actuators.

An engineering prototype with a 62 cm thin meniscus mirror to investigate the practical problems of the JNLT active optics system has been constructed. The structure and the performance of the prototype and the basis for the control algorithms are described. The stiffness of the mirror and the purity of figure control for the correction of lower order Zernike modes were measured for this system. The results of experiments to control the surface figure of the mirror under varying elevation angle of the telescope or under external force load are reported. The thermal effects on this system were also evaluated.

The design of aluminization systems for the MMT Conversion 6.5 m mirror and the Columbus Project 8 m mirror has led us to reconsider many of the design issues and tradeoffs for such systems. Coating of the large honeycomb mirrors will be done in situ on the telescope with a portable bell jar forming the front half of a two-stage vacuum system. The mirror cell forms a "dirty" vacuum behind the mirror to eliminate excess force on the glass. A multi-ring source geometry has been proposed to allow a 1.0 m spacing between the mirror surface and the sources thereby minimizing the size of the vacuum chamber. Evaporation source models have been developed to optimize the number of sources, ring spacing, and high incidence angle emission to achieve better than 5% rms deviation in coating thickness over the diameter. Code results are compared to empirical thickness profiles measured at the University of Arizona's (UA) Sunnyside 2.0 m coating facility. Cryoadsorption pumps are considered for reasons of economy, quality of vacuum, pumping speed, and reliability. The interaction of the cryopumps and getter pumping with the pumping/cleaning/deposition cycle is studied. Glow discharge cleaning is discussed and the results of deposition tests in 10' Torr residual argon are given. Electrical requirements are estimated and a novel transformer design may decrease the current entering the chamber from 12,000 A to less than 600 A.

A method of phasing segmented-mirror systems using a new form of white-light radial-shear interferometer is presented. The device was used in a laboratory proof-of-concept demonstration to phase two elements of an annularly segmented mirror. This co-phasing was achieved both manually and with low-bandwidth closed-loop servo control using a new and perhaps simpler modulation technique than that proposed previously. The configuration used in the demonstration is described, together with some initial results.

The pistons and tilts of the 36 segments of the Keck Telescope primary mirror are under active control. The mechanical and electronic designs of the actuators used to achieve this control are described along with the performance of the actuators under a variety of tests. In use, the actuators will move in 4-nm increments. This resolution and the accuracy of the actuator moves are adequate for stabilizing the figure of the primary mirror to the precision required for optical and infrared astronomy.

The new Phased Array Mirror, Extendible Large Aperture (PAMELA) technology can be applied to innovative designs for large astronomical telescopes and laser beam directors. Filled primary apertures with diameters exceeding 15 m will be capable of diffraction-limited imaging of objects at visible wavelengths. The same aperture can function as a nearly perfect conventional telescope primary mirror when objects or reference stars are too faint for adaptive compensation of atmospheric turbulence. This performance will be accomplished by means of a dual-mode control system which utilizes either local figure sensing to yield a nearly perfect wide-field, highly segmented mirror or a fully adaptive wavefront compensation system to drive the mirror segments to the phase conjugate positions to correct for atmospheric turbulence within the isoplanatic angle restriction. The active optics sensing for local figure control can be accomplished either by using differential height and tilt sensors on each segment or by using an optical interferometer at the center of curvature of the primary mirror. This paper presents a design concept for a 10-m telescope with a spherical primary mirror composed of about 10,000 10-cm diameter hexagonal segments. Principles of operation are described, and estimates of performance are given.

The ten meter diameter primary mirror of the W. M. Keck Telescope is a mosaic of thirty-six hexagonal mirrors. An active control system stabilizes the primary mirror. The active control system uses 168 measurements of the relative positions of adjacent mirror segments and 3 measurements of the'primary mirror position in the telescope structure to control the 108 degrees of freedom needed to stabilize the figure and position of the primary mirror. The components of the active control system are relative position sensors, electronics, computers, actuators that position the mirrors, and software. The software algorithms control the primary mirror, perform star image stacking, emulate the segments, store and fit calibration data, and locate hardware defects.

The Primary Mirror of the Keck Observatory Telescope is made up of an array of 36 hexagonal mirror segments under active control. The measurement of the relative orientations of the mirror segments is fundamental to their control. The mechanical and electronic design of the sensors used to measure these relative positions is described along with the performance of the sensors under a variety of tests. In use, the sensors will measure relative positions with a resolution of a few nanometers. This resolution and the low noise, drift and thermal sensitivity of the sensors are adequate to stabilize the primary mirror figure to the precision require for optical and infrared astronomy.

The active control system (ACS) uses both parallel and distributed processing techniques to measure and control the posi- Lions of the 36 segments of the Keck Observatory Telescope primary mirror. The main function of the software is to mainthin the mirror figure; to accomplish this goal the software uses a predictive, "feed-forward" mechanism which effectively increases the system bandwidth for the most important sources of perturbation. The software executes on a set of twelve 68000-family processors under the supervision of a VAX workstation. An array of nine parallel I/O processors collect and process data from 168 displacement sensors and transmit motion commands to 108 actuators. Three additional processors simultaneously compute actuator commands, monitor system performance, compute sensor control parameters and communicate with other observatory computers. The software is highly optimized for speed.

The pistons and tilts of the 36 segments of the Keck Telescope primary mirror are under active control. The mechanical and electronic designs of the actuators used to achieve this control are described along with the performance of the actuators under a variety of tests. In use, the actuators will move in 4-nm increments. This resolution and the accuracy of the actuator moves are adequate for stabilizing the figure of the primary mirror to the precision required for optical and infrared astronomy.

The Primary Mirror of the Keck Observatory Telescope is made up of an array of 36 hexagonal mirror segments under active control. The measurement of the relative orientations of the mirror segments is fundamental to their control. The mechanical and electronic design of the sensors used to measure these relative positions is described along with the performance of the sensors under a variety of tests. In use, the sensors will measure relative positions with a resolution of a few nanometers. This resolution and the low noise, drift and thermal sensitivity of the sensors are adequate to stabilize the primary mirror figure to the precision require for optical and infrared astronomy.

The Primary Mirror Active Control System (ACS) of the W.M. Keck Telescope has as its main function the maintenance of the mirror figure of the 36-segment primary mirror under the changing effects of gravity, temperature, and other low frequency perturbations. The ACS is a multivariate control loop that can be represented in a diagonalized form, provided that segment actuator motions only excite oscillations in its corresponding whiffletree (tying an actuator to a mirror segment), with no coupling to the other whiffletrees in the same segment mirror. Since whiffletree oscillations are expected to occur at frequencies above the bandpass of the control system, the assumption is expected to be valid for the purpose of analyzing the stability and response of the ACS under the expected low frequency perturbations. The results of a one-dimensional simulation, justified by the diagonalized form of the problem, will be presented showing the conditions for stability, the system response to desired changes and the advantages of using feed-forward. A verification of the theoretical results will be presented for an actual actuator coupled to a sensor controlled by a one-dimensional version of the ACS software. Also based on the diagonalized form, a study of noise coupling, equivalent system bandwidth and matrix noise magnification factor will be presented. The effect of the feed-back control loop on the telescope mis image radius caused by sensor noise will be calculated.

An alt-alt (altitude-altitude) mounting for an 8-metre telescope has several operational advantages over an alt-az (altitude-azimuth) design. In this alt-alt arrangement the yoke (or major) axis, which is horizontal, lies East-West and the second (or minor) altitude axis is located across the yoke and lies in the North-South plane. By comparison, the alt-az has a vertical, azimuth axis pointed at the zenith and its minor axis is always horizontal. Consequently, the altaz telescope cannot track through the zenith because of the infinitely high rotation rate required about the vertical (azimuth) axis, thus it cannot be used to observe at the zenith where the images are sharpest and the atmospheric transmission highest, and its axle and field rotation speeds are disadvantageously high throughout a considerable area of sky in the vicinity of the zenith. The rapid rotation of spectrographs at folded Cassegrain and Nasmyth foci can cause the spectrum to drift on the detector during an exposure because of fiexure within the spectrographs. By comparison, the alt-alt telescope tracks readily through the zenith at slow rotation speeds, and the field rotation rate is zero when tracking along the Celestial Equator where the telescope becomes pseudo-equatorial because one axis (the minor axis) is then parallel to the axis of the Earth. It is at the Eastern and Western horizons where the rate of rotation about the horizontal axis becomes excessive, which is not critical because telescopes are not scheduled to observe there, where the seeing and transparency are very poor, and where in practice the light is usually obstructed by the building enclosing the telescope. When pointed at the zenith, the primary mirror in an alt-alt mounting is suspended over the cooled observing floor, thus minimizing seeing degradation caused by heat generated in the bearings and drives. (It is when the telescope is pointed closer to the horizon that the light path within the telescope tube is over a bearing.) Because of the rapid rotation and acceleration during tracking the alt-az mountings are specified to have exceptionally high resonant frequencies to enable quick response to the drive motors. The alt-alt telescope does not need such high resonance frequencies because its rotation speeds are low, usually even lower than for an equatorial mounting. In summary, the characteristics of the alt-alt mounting are in phase with the astronomical requirements, whereas the alt-az mounting is 90 degrees out of phase with these requirements in the sense that its worst performance is at the zenith instead of at the horizon.

A technique has been developed for casting glass in the form of a honeycomb structure possessing good stiffness despite its low weight, and facilitating thermal control via forced ventilation of the honeycomb cells. To date, 3.5 m diameter mirrors of this type have been successfully cast; fully 8 m diameter mirrors are expected to be cast by 1992. No mirror of diameter as large as 8 m has ever been polished, however, and the difficulties which will be encountered shall be compounded by the shorter focal lengths required for advanced telescopic optics. A novel method, designated 'stressed lap polishing', has been developed to address these problems.

Recently, design examples of wide field three mirror Mersenne-Schmidt system had been presented by R.V. Willstrop. The work presented in this paper is an extension of Willstrop's early work. By applying overall system optimization, the maximum image size within a 4-degree field of a well-designed Mersenne-Schmidt system can be as small as 0.16 arc sec. If small edge vignetting is allowed, within a five-degree wider field of view, the maximum image size of the Mersenne-Schmidt system would be as small as 0.34 arc sec. In the second part of this paper, the Mersenne-Schmidt telescope is discussed in comparison with a Schmidt camera. Discussion shows that both systems can achieve a wider field of view. However, from astronomical requirements, the Mersenne-Schmidt telescope has better image quality, wider waveband coverage, greater light power, shorter tube length and larger usage area of the primary mirror. For a light collecting area larger than that of existing ones, the Mersenne-Schmidt telescope would be a better choice for astronomical frontier work.

The Central-Axis Reflector (ZAS), the design of which is presented, is a segmented mirror telescope. The inventions relate mainly to the optical system and to the tracking apparatus. A number of individual mirror bodies, ground off-axis (hexagonal/polygonal) produce one primary mirror with closed circular aperture when joined together. The overall design of the tracking apparatus results directly (and thus without unnecessary adornment) from the two planes of motion which have been reduced to an obligatory minimum but which are required for tracking of the telescope.

A system aimed at real-time wavefront analysis at the focus of an astronomical telescope has been developed. The method uses Zernike polynomial descriptions of the fringes from a lateral shearing interferometer and the resulting Zernike coefficients for the sheared state are matrix transformed to those of the real wavefront. Tests to confirm that the method operates correctly are described and preliminary results are given with the interferometer at the focus of a Ritchey-Chretien telescope. Applications include quantitative evaluation and correction of telescope mirror misalignments, active optics, and the determination of dome and atmospheric seeing parameters.

The Stratospheric Observatory For Infrared Astronomy (SOFIA) currently under development will mount a 2.5-m aperture IR telescope on an airborne (B 747 SP) platform for observations, at altitudes in excess of 12.5 km, in the sub-mm to visible spectral range. The first part of this work discusses the design tradeoffs which arose during the predevelopment phase in such considerations as primary mirror size and overall configuration. The second part gives attention to the incorporation of such advanced technology features as the thin Zerodur meniscus mirror concept used for the six-mirror compound telescope configuration, and the CFRP materials of the support structure.

The theory, methods, requirements, limitations, effectiveness and economics of cleaning large telescope mirrors with carbon dioxide (COJ snowflakes are discussed. The method holds promise as a rapid, easy-to-use technique for routinely maintaining the cleanliness of mirrors between washings.

Normally the atmosphere dispersion compensator (ADC) is a permanent fixture in the image corrector, even though at times the ADC is not needed. But, in Cassegrain focus it is possible to design a removable ADC, which can be taken out to achieve higher light transmission. Three such design examples are given in this paper. All of them are designed for the Cassegrain focus of the planned NOAO 8-meter telescope project. At this focus, the aberration corrector has three lenses. In the first two designs, the ADC has to be inserted between the second and third lens of the corrector; in the third the ADC is placed behind the third lens of the corrector. The first design requires adjustment of the secondary mirror of the telescope as well as the third lens of the aberration corrector when the ADC is inserted. The second design does not require any adjustment but requires two extra curved surfaces in the ADC. In the third design, the focal surface, the secondary mirror, and the third lens require adjustment. Spot diagrams of the three designs at three zenith distances are given.

The design for the Columbus binocular telescope has been completed and the project is now proceeding to the construction phase. The two 8-m f/1.2 borosilicate honeycomb mirrors will give an effective collecting area of 11.3 m and a baseline of 22 m for interferometry. The optics are supported on an exceedingly compact and rigid 'C-ring' mount with a lowest eigenfrequency of around 10 Hz. The system will be housed in a corotating enclosure and will be located on Mt. Graham in southeastern Arizona. Instrumentation priorities have been established for the telescope and conceptual designs are now under development.

Optical communications using laser light in the visible spectral range is being considered for future deep-space missions. Such a system will require a large telescope in earth vicinity to be used as a receiving station for data return from the spacecraft. A preliminary discussion for a ground-based receiving station consisting of a 10-meter hexagonally segmented primary with high surface tolerance and a unique sunshade is presented.

The Vatican Advanced Technology Telescope (VATT) now being fabricated differs from traditional telescopes in many ways. The altitude over azimuth mount will be direct driven by large diameter motors. The cell for the f/1 borosilicate honeycomb primary mirror incorporates a thermal control system to stabilize the mirror temperature. The f/9 Gregorian secondary will be mounted on a six-axis stage and controlled to submicron resolution in order to maintain the strict collimation tolerances needed for the fast optical system. Though only 1.83 m in aperture, VATT incorporates many of the design features of larger projects in the 8 m class.

A computer-controlled optical surface (CCOS) process has been developed that is in routine use for fabricating off-axis and centered aspheric mirrors. The CCOS process effects surface removal by moving a relatively small tool over the mirror surface in a path covering the entire surface. The removal is computed by the convolution of the tool work function with the path of the tool over the mirror surface. The combination of CCOS with microgrinding (grinding with fine tool over the mirror surface. The combination of CCOS with microgrinding (grinding with fine diamond powders which produce a specular surface) allows interferometric testing at an early stage of the process. Surface figure accuracies better than 20 nm rms and finishes better than 10 A are currently achieved.

In cooperation with NASA, the West German Minister for Research and Technology (BMFT) since 1986 has conducted studies to provide an airborne telescope of 2.5 m aperture for SOFIA, a successor to the extremely successful Gerard Kuiper Airborne Observatory (KAO). In the decades from the mid-nineties onward, SOFIA could make available a continuing and readily deployable observation platform at heights above 12.5 km for sophisticated instrumentation from the submm well into the optical spectral range. While the planned SOFIA facility is described in part I, here an overview will be given of the technical problems involved in developing a large aperture optical telescope for operation in an open cavity environment of an aircraft flying at high altitude. Issues that have been addressed in the project definition phase comprise the figuring and handling of a 2.7 m thin meniscus mirror with very fast f/ratio 1.2, the novel support system, the design of the telescope assembly using light weight composites, the large spherical bearing and the telescope control and drive system. Some of the aforementioned technicalities are being studied on a subscale level, others still have to mature until the beginning of the development phase.

New opportunities and new difficulties arise from technological advances in modes of astronomical observing and data transmission. This paper gives a quick re-cap of recent experiences in operations of space observatories as seen from the perspective of a practicing operator. Lessons learned from different approaches to design and execution of telescope construction and control are presented relating to their actual use in practice. Suggestions for better ways to use future telescopes and to maximize their operational efficiency will be offered. The emphasis will be on considering operational aspects in the planning stages. This will require technical interaction among the scientists planning to use advanced technology telescopes for obtaining efficient receipt of high quality data, the engineers involved in developing the hardware and operating systems, and the operators who will directly interact with the advanced technology to accomplish this goal.

Normally moderate or large aperture telescopes are designed and constructed on a custom, limitednumber basis. This fact alone makes acquisition of the optics a long term project. Telescopes have been produced in volume in the past but only in relatively small apertures. Recent developments with robotic telescopes has stimulated interest in the possibility of a Global Network of Automatic Telescopes (GNAT). Although large volume production of most of the elements of this technology is possible, production problems of conventional two-mirror optical designs in the 1-meter class would greatly handicap such a venture. To meet both the needs of high-volume production and overall optical performance, a new fourmirror optical design is required.

We describe a new spectroscopic facility based upon a novel Multi- Telescope Telescope (MTT) and a fiber optic fed spectrograph. The MTT is an inexpensive one meter light collecting telescope, whose "primary" mirror consists of nine commercially made amateur-size (33.3 cm) telescope mirrors. Each mirror of the MTT will focus light into a separate optical fiber, thus avoiding the light loss associated with the dead space in conventional fiber bundles. Throughput is further enhanced by low mirror obscuration and having only one reflection from the high reflectivity coatings available for smaller mirrors. The telescope is thus equivalent to about a 1.3-m conventional telescope for spectroscopy. The optical fibers will feed an off-plane Ebert-Fastie spectrograph with a COD detector. We conservatively estimate an overall optical efficiency of about 2.7% (4.4% without a slit at somewhat lower resolution), which corresponds to a limitirg magnitude of 9.4 for 1000 sec exposure, SNR=100, and a resolution of 0.2 A/pixel. By using a combination of an innovative structure, a control system, some existing materials, and donated shop time we can reduce the material cost of the telescope, spectrograph, and detector to about *35K to $55K, depending on the detector. The MTT and spectrograph will be installed at the newly commissioned Georgia State University Hard Labor Creek Observatory and will be used for spectroscopic observations of binaries and nonradial pulsations in Be stars.

One of the major design guidelines of the CHARON program (consisting of a cross shaped array of five 26 cm diameter telescopes allowing variable baselines up to 50 m) was to avoid the utilization of a laser metrology, principally for simplicity and cost reasons. The objective was to achieve a delay line whose vibration level during translation remains negligible, i.e., less tham 12 nm RMS for a 15 msec exposure time, without any measurement of the actual optical path length introduced by the delay line. The optical delay line is built around a 1-meter stroke, retail-market translation table whose carriage can move at speeds between 0.1 micron/sec and 1 mm/sec, while generating a low vibrational level. Results indicate that the performance values achieved were satisfactory for both optical and mechanical aspects.

An adaptive optics prototype system has been tested at the 1.52 m telescope of the Observatoire de Haute Provence, resulting in diffraction-limited images at near infrared wavelengths (2.2 to 5 microns). This paper presents the first results and a short analysis, which demonstrate the considerable gain in resolution and sensitivity achieved by this technique. Single stars, close binary stars, and a satellite have been resolved. In some cases another star several arcseconds apart has been used as reference for the wavefront sensing.

Results recently obtained for the use of the curvature-sensing method as a substitute for slope sensing in optical wavefront reconstruction, using long-exposure CCD images of the beam cross-section on either side of the telescope focal plane. A program based on the solution to the Poisson equation is then applied in order to reconstruct the wavefront. Relative to the existing Hartmann sensing methods, curvature-sensing yields sensitivity comparable to that of the Shack-Hartmann test. Additional optics and reference plane-based calibration are obviated. Tests of the new method on an 88-inch Ritchey-Chretien telescope have yielded a map of residual wavefront errors as a solution of the Poisson equation.